Sweet spot prediction method for superpressure and low-permeability gas reservoir based on grain size reconstruction seepage evaluation model

Abstract

This invention provides a method for predicting sweet spots in overpressured, low-permeability gas reservoirs based on a grain size reconstruction seepage evaluation model. This method uses core grain size analysis data and conventional logging curves, employing a neural network algorithm to train a core grain size prediction model. Using the Kozeny-Carman equation and the fitting coefficients of core mercury injection data, a conversion model between core grain size and pore throat radius is obtained. By defining a natural seepage capacity index, an energy factor, and a comprehensive natural sweet spot index, the conversion model is reconstructed into a novel evaluation model coupling seepage capacity and formation energy attributes. Based on this evaluation model, and combined with highly sensitive 3D seismic attributes related to porosity, permeability, and gas content, a 3D evaluation model of the comprehensive natural sweet spot index is output. The beneficial effects of this invention are: it can intuitively, accurately, and cost-effectively predict the "natural sweet spot" of low-permeability gas reservoirs, and quantitatively evaluate the natural production potential of the reservoir.

Description

Technical Field

[0001] This invention relates to the field of oil and gas field exploration and development technology, and more specifically to a method for predicting sweet spots in overpressured and low-permeability gas reservoirs based on a grain size reconstruction seepage evaluation model. Background Technology

[0002] Offshore overpressured and low-permeability gas reservoirs represent an important successor area for oil and gas development, but their development faces enormous challenges. Due to extremely poor reservoir properties, strong heterogeneity, and the presence of overpressure environments, conventional logging responses are insufficient for identifying effective reservoirs, making the prediction of "sweet spots" extremely difficult.

[0003] Existing technologies mainly include: 1. Taking Chinese patent document CN120103476A as an example, CN120103476A performs high-frequency attenuation gradient attribute processing, structural smoothing processing, and extraction of maximum curvature attribute and ant body attribute on pre-stack CRP gather data, pre-stack migrated pure wave seismic data and post-stack result data, and forms reservoir planar development characteristics and distribution patterns, and classifies and zons them, and predicts reservoir sweet spots.

[0004] 2. Taking Chinese patent document CN120429558A as an example, CN120429558A uses post-stack seismic data and, based on the GR inversion model and the gradient value of the time window wave spectrum ratio attenuation, selects the target sweet spot region.

[0005] 3. Taking Chinese patent document CN113945970B as an example, CN113945970B combines the wave impedance inversion volume with the instantaneous frequency attribute volume. Based on the frequency variation attribute difference method of matched pursuit time-frequency analysis, the difference volume of two dominant frequency single-frequency volumes is obtained. Then, the dense sandstone sweet spot attribute volume is obtained by combining according to the formula. By identifying the area where the average value of the sweet spot attribute volume is greater than the set value, the development area of ​​the dense sandstone sweet spot is predicted.

[0006] In summary, current sweet spot prediction methods for such reservoirs tend to focus on "fractureable reservoirs," that is, on the brittleness and mechanical properties of the rock to provide targets for hydraulic fracturing. However, large-scale fracturing is not feasible in offshore environments. Under these "non-fracture" development models, there is an urgent need for a method to accurately identify "natural sweet spots" within the reservoir that possess inherent industrial flow capabilities. Existing technologies lack a detailed characterization of the coupling relationship between the microscopic pore structure controlling the natural flowability of natural gas and formation energy, resulting in low drilling success rates and unsatisfactory development outcomes. Summary of the Invention

[0007] This invention overcomes the shortcomings of the prior art and provides a method for predicting sweet spots in overpressured, low-permeability gas reservoirs based on a particle size reconstructed seepage evaluation model.

[0008] A method for predicting sweet spots in overpressured, low-permeability gas reservoirs based on a particle size reconstructed seepage evaluation model includes the following steps: S1. Using a neural network algorithm, a multivariate nonlinear relationship is fitted based on core grain size and well logging curve data to obtain a core grain size curve prediction model. Through the core grain size curve prediction model, a grain size median curve prediction model and a sorting coefficient curve prediction model are obtained. S2. Substitute the core grain size curve prediction model, grain size median curve prediction model, and sorting coefficient curve prediction model obtained in S1 into the conversion relationship model between grain size parameters and pore throat radius. Through the conversion relationship model between grain size parameters and pore throat radius, calculate the equivalent pore throat radius curve model. S3. Define the Natural Permeability Index (NFCI) and obtain the NFCI definition formula. S4. Substitute the definition formula of the natural seepage capacity index NFCI in S3 into the equivalent pore throat radius curve model obtained in S2, calculate the natural seepage capacity index NFCI, and calculate the equivalent pore throat radius curve model as the natural seepage capacity index curve model. S5. Using gas production capacity data from gas testing or core samples, determine the industrial lower limits of the natural permeability index NFCI and porosity Φ, NFCI_min and Φ_min. On a single well profile, identify all segments that simultaneously satisfy NFCI > NFCI_min and Φ > Φ_min, define the identified segments as "natural sweet spots", and calculate the cumulative thickness h_net of the "natural sweet spots". S6. Define the energy factor EF and obtain the formula for defining the energy factor EF; construct the NSSCI curve model of the natural dessert comprehensive index based on the formula for defining the energy factor EF. S7. Extract the NFCI values ​​from the natural permeability index curve model point by point according to depth. Based on the NFCI_min and Φ_min values ​​determined in S5, select the depth points in the "natural sweet spot layer". Substitute the selected depth points into the natural sweet spot comprehensive index NSSCI curve model to obtain the NSSCI curve model that couples permeability and formation energy properties. S8. Using the natural sweet spot index (NSSCI) curves of each single well as hard data, three-dimensional seismic attributes sensitive to porosity, permeability, and gas content are simultaneously selected as auxiliary constraint data. Geostatistical methods are used to establish a three-dimensional NSSCI attribute volume. In the three-dimensional NSSCI attribute volume, continuous spatial regions with NSSCI values ​​greater than a set threshold are extracted, and the obtained continuous spatial regions are defined as "sweet spot connected bodies".

[0009] The conversion model between particle size parameters and pore throat radii in S2 is as follows: r_equiv = a * (D50)^b * (Sort)^c * Φ^d (1) Where a, b, c, and d are coefficients obtained by fitting the core mercury intrusion data, D50 is the grain size median curve, and Sort is the sorting coefficient curve; The specific formula for defining the Natural Permeability Index (NFCI) is as follows: NFCI = (K * r_equiv) / (μ_g * B_g) (2) Where K is the permeability calculated from the particle size or pore throat parameters, μ_g is the formation natural gas viscosity after PVT correction, and B_g is the natural gas volume factor.

[0010] The energy factor EF is defined by the following formula: EF = (P_p - P_wf) / P_wf (3) Where P_p is the formation pore pressure and P_wf is the reasonable bottom flow pressure of the production well.

[0011] The formula for calculating the Natural Sweetness Index (NSSCI) is as follows: NSSCI = NFCI * EF * (1 - S_wi) * h_net (4) Where S_wi is the bound water saturation and h_net is the cumulative thickness of the "natural dessert layer".

[0012] S1 uses software to execute the neural network algorithm, and the execution software is PyCharm.

[0013] The beneficial effects of this invention are as follows: This invention is applicable to highly heterogeneous, overpressured, low-permeability gas reservoirs. Based on analytical data and conventional logging curves, a multi-parameter prediction model is formed using formulaic methods and machine learning. By defining a natural permeability index, energy factor, and natural sweet spot index, the conversion model is reconstructed into a novel evaluation model coupling permeability and formation energy attributes. Based on this evaluation model, combined with highly sensitive three-dimensional seismic attributes related to porosity, permeability, and gas content, a three-dimensional evaluation model of the natural sweet spot index is output. This model can intuitively, accurately, and cost-effectively predict the "natural sweet spot" of low-permeability gas reservoirs and quantitatively evaluate the natural production potential of the reservoir. Attached Figure Description

[0014] Figure 1 This is a flowchart illustrating the method. Figure 2 This is a graph of the curve prediction model; Figure 3 A bar chart for identifying single wells in the natural sweet spot layer; Figure 4A 3D seismic / NSSCI attribute volume diagram; Figure 5 This is a connected diagram of natural desserts. Detailed Implementation

[0015] Example like Figure 1 As shown, the method for predicting sweet spots in overpressured, low-permeability gas reservoirs based on a particle size reconstruction seepage evaluation model includes the following steps: S1. Using a neural network algorithm, a multivariate nonlinear relationship is fitted based on core grain size and well logging curve data to obtain a core grain size curve prediction model. Through the core grain size curve prediction model, a grain size median curve prediction model and a sorting coefficient curve prediction model are obtained.

[0016] Step S1 involves fitting a multivariate nonlinear relationship based on core grain size analysis data and conventional logging curves.

[0017] Preferably, the core particle size data in S1 is core particle size analysis data processed by a laser particle size analyzer. The laser particle size analyzer calculates the core particle size data, including the particle size, median particle size, and sorting coefficient of each sample, based on the angle and intensity characteristics of the scattered light, combined with optical models and algorithms, by irradiating the dispersed core sample particles with a laser.

[0018] Preferably, the logging curve is a curve that is sensitive to changes in core grain size, such as GR, AC, DEN, CNL, and RT logging curves.

[0019] Preferably, the prediction model for the core grain size, median grain size, and sorting coefficient curves is a curve prediction model obtained by performing multivariate nonlinear fitting of the experimental analysis scatter data of core grain size, median grain size, and sorting coefficient with continuous logging curves.

[0020] Furthermore, S1 employs software to execute the neural network algorithm, using PyCharm software. PyCharm is a professional Python integrated development environment (IDE) that can be used for code debugging, data science, and machine learning. It can call core grain size analysis experimental data and well logging curve data and use novel neural network algorithms to obtain various curve prediction models.

[0021] Neural network algorithms simplify the hidden layer structure of traditional neural network algorithms through pruning techniques, sparsification, and neuron elimination. At the same time, attention mechanisms are embedded in the algorithm based on the results of the main geological factors. By focusing on key input features through multi-weighted methods, the model's ability to capture and utilize important information is improved.

[0022] S2. Substitute the core grain size curve prediction model, grain size median curve prediction model, and sorting coefficient curve prediction model obtained in S1 into the conversion relationship model between grain size parameters and pore throat radius. Through the conversion relationship model between grain size parameters and pore throat radius, calculate the equivalent pore throat radius curve model.

[0023] The conversion model between particle size parameters and pore throat radii in S2 is as follows: r_equiv = a * (D50)^b * (Sort)^c * Φ^d (1) Where a, b, c, and d are coefficients obtained by fitting mercury intrusion porosimetry data from the core, D50 is the median grain size curve, and Sort is the sorting coefficient curve.

[0024] In S2, the conversion model between grain size parameters and pore throat radius is based on core mercury intrusion porosimetry data and utilizes a specialized modification of the Kozeny-Carman equation to establish the conversion equation between grain size parameters and pore throat radius. The Kozeny-Carman equation is a classic empirical equation describing the quantitative relationship between permeability and pore structure parameters, and its core function is to calculate the difficult-to-measure permeability using easily measurable pore characteristic parameters.

[0025] Core mercury injection data, with the relationship between injection pressure and mercury saturation in the core pores as the core, can intuitively and quantitatively reflect the pore size distribution, throat connectivity and pore morphology of the core pore system.

[0026] Preferably, the equivalent pore throat radius curve model, core grain size curve prediction model, grain size median curve prediction model, and sorting coefficient curve prediction model are all calculated using PyCharm software to obtain the final curve model.

[0027] S3. Define the Natural Permeability Index (NFCI) and obtain the NFCI definition formula.

[0028] The specific formula for defining the Natural Permeability Index (NFCI) is as follows: NFCI = (K * r_equiv) / (μ_g * B_g) (2) Where K is the permeability calculated from the particle size or pore throat parameters, μ_g is the formation natural gas viscosity after PVT correction, and B_g is the natural gas volume factor.

[0029] The Natural Permeability Index (NFCI) is a quantitative index that expresses the flow capacity and characteristics of natural gas in a reservoir. It is calculated using formulas based on permeability calculated from particle size or pore throat parameters, formation natural gas viscosity corrected by PVT, natural gas volume factor, and an equivalent pore throat radius curve model. PVT-corrected natural gas viscosity is obtained by systematically correcting the original calculated or measured viscosity values ​​by incorporating key PVT parameters such as pressure, temperature, composition, and phase state. This eliminates deviations caused by differences in operating conditions and accurately reflects the viscous characteristics of natural gas under actual reservoir or transportation conditions. PVT, an abbreviation for Pressure, Volume, and Temperature, is a discipline or parameter system that studies the physical properties (such as density, viscosity, and phase state) and their changes of substances (especially fluids like oil and gas) under different pressure, volume, and temperature conditions. Its related data can be used to correct calculated fluid properties, ensuring they conform to actual operating conditions and providing accurate basic parameters for oil and gas reservoir development, fluid transportation, and other engineering projects. The natural gas volume factor refers to the ratio of the volume occupied by a unit mass of natural gas under a certain actual pressure and temperature condition (such as reservoir or transportation conditions) to the volume occupied by a unit mass of natural gas under standard conditions (usually 0℃, 101.325kPa). It essentially reflects the compression or expansion characteristics of natural gas as pressure and temperature change, and is an indispensable basic parameter in engineering such as oil and gas reservoir reserve calculation, production accounting, and fluid seepage simulation.

[0030] S4. Substitute the definition formula of the natural seepage capacity index NFCI in S3 into the equivalent pore throat radius curve model obtained in S2, calculate the natural seepage capacity index NFCI, and convert the equivalent pore throat radius curve model into a natural seepage capacity index curve model.

[0031] S5. Using gas production capacity data from gas testing or core samples, determine the industrial lower limits of the natural permeability index NFCI and porosity Φ, NFCI_min and Φ_min. On a single well profile, identify all segments that simultaneously satisfy NFCI > NFCI_min and Φ > Φ_min, define the identified segments as "natural sweet spots", and calculate the cumulative thickness h_net of the "natural sweet spots".

[0032] Among them, gas testing is a key project carried out in the exploration and development stage of oil and gas fields for the target reservoir. After reservoir stimulation measures such as perforation and acidizing / fracturing, fluid drainage, production capacity testing and pressure monitoring are carried out to obtain core data such as reservoir production, pressure changes and fluid properties.

[0033] Core gas production capacity data is a comprehensive set of data obtained by conducting analytical and flow tests on core samples from oil and gas reservoirs. It includes key parameters such as gas production rate, cumulative gas production, and gas saturation. The core of this data is used to evaluate the gas production potential of the corresponding reservoir.

[0034] The industrial lower limit values ​​of the Natural Permeability Index (NFCI) and porosity Φ are used to determine whether the natural gas in a single well has industrial flow capacity or economic benefits within a certain stratum. When NFCI and Φ are below the industrial lower limit values, the natural gas in that stratum has industrial flow capacity or economic benefits.

[0035] Industrial flow capacity refers to the ability of a reservoir, under actual development conditions, for fluids (oil, gas, water) to flow stably at a rate that meets industrial extraction standards, taking into account factors such as reservoir permeability, fluid viscosity, and formation pressure differences.

[0036] The thickness of the natural sweet spot layer h_net refers to the thickness of the layer within the target layer where NFCI and Φ are higher than the industrial lower limit. The natural gas within the aforementioned layer has industrial flow capacity or economic benefits and is a natural sweet spot that has not been artificially modified.

[0037] S6. Define the energy factor EF and obtain the formula for defining the energy factor EF; construct the NSSCI curve model of the natural dessert comprehensive index based on the formula for defining the energy factor EF.

[0038] The energy factor EF is defined by the following formula: EF = (P_p - P_wf) / P_wf (3) Where P_p is the formation pore pressure and P_wf is the reasonable bottom flow pressure of the production well.

[0039] The formula for calculating the Natural Sweetness Index (NSSCI) is as follows: NSSCI = NFCI * EF * (1 - S_wi) * h_net (4) Where S_wi is the bound water saturation and h_net is the cumulative thickness of the "natural dessert layer".

[0040] Among them, the energy factor is a quantitative characterization of the natural gas seepage capacity of the corresponding formation by calculating the pressure difference between the formation pore pressure and the reasonable bottom hole flowing pressure and the ratio of the pressure difference to the bottom hole flowing pressure.

[0041] Formation pore pressure refers to the fluid pressure exerted within the pore spaces of underground rocks where natural gas is stored. Essentially, it is the pressure acting within the reservoir pores, and it is the primary driving force propelling natural gas from the underground reservoir to the wellbore.

[0042] Optimal bottomhole flowing pressure refers to the optimal pressure value controlled at the bottom of the wellbore during gas well production. This pressure must ensure a continuous and stable flow of gas from the formation into the wellbore while effectively preventing damage such as sand production, water-gas coning, or physical destruction of reservoir rocks caused by excessively low pressure. It balances the conflict between maximizing production and ensuring long-term stable gas production, ultimately achieving the highest final recovery rate and best economic benefits for the gas reservoir.

[0043] The Natural Sweet Spot Composite Index (NSSCI) is a predictive model calculated using software with defined formulas based on four parameters: Natural Permeability Index (NFCI), Energy Factor (EF), Bound Water Saturation (S_wi), and Natural Sweet Spot Layer Thickness (h_net). This composite index simultaneously couples the natural gas permeability of the target layer with formation energy properties, and can effectively evaluate the relevant parameters of natural gas production flow in a specific natural sweet spot layer.

[0044] S7. Extract the NFCI values ​​from the natural permeability index curve model point by point according to depth. Based on the NFCI_min and Φ_min values ​​determined in S5, select the depth points in the "natural sweet spot layer". Substitute the selected depth points into the natural sweet spot comprehensive index NSSCI curve model to obtain the NSSCI curve model that couples permeability and formation energy properties.

[0045] S8. Using the Natural Sweet Spot Index (NSSCI) curves of each individual well as hard data, and simultaneously selecting three-dimensional seismic attributes sensitive to porosity, permeability, and gas content as auxiliary constraint data, a three-dimensional NSSCI attribute volume is established using geostatistical methods. Within this volume, continuous spatial regions with NSSCI values ​​greater than a set threshold are extracted and defined as "sweet spot connected bodies." After obtaining these "sweet spot connected bodies," the well location and trajectory design are optimized based on their spatial distribution to ensure that drilling maximizes the encounter with the sweet spot connected bodies and vertically extending natural sweet spot layers.

[0046] Preferably, in S8, continuous spatial regions with NSSCI values ​​greater than a set threshold are extracted. The set threshold is calculated based on the NFCI and Φ values ​​of different gas fields.

[0047] Among them, three-dimensional seismic attributes refer to mathematical parameters extracted from three-dimensional seismic data volumes to quantitatively describe the structure and physical properties of underground rock strata. Through mathematical transformations, it enhances and decomposes information such as the amplitude, frequency, and phase of seismic waves, generating various data volumes that can indirectly reflect changes in lithology, physical properties, and fluids.

[0048] Geostatistics is a mathematical prediction technique based on regionalized variable theory and spatial correlation. It quantitatively characterizes the spatial structure and variability of geological parameters (such as porosity and grade) through variograms and utilizes algorithms such as Kriging to perform optimal unbiased estimations of unsampled points. The core of this method lies in acknowledging and utilizing the geological principle that "closer distances lead to more similar attributes," thereby enabling scientific predictions from discrete well data to three-dimensional geological models.

[0049] The three-dimensional NSSCI attribute volume is established using stochastic simulation methods (such as sequential Gaussian simulation) and incorporating three-dimensional seismic attribute constraints. The difference between the three-dimensional NSSCI attribute volume and the NSSCI curve model is that the three-dimensional NSSCI attribute volume is more intuitive and three-dimensional, and has a stronger ability to represent the spatial distribution of natural desserts.

[0050] Stochastic simulation is a technique that characterizes the spatial distribution uncertainty of underground reservoir properties by generating multiple statistically equally likely high-resolution implementations that are consistent with known sample data and spatial structure.

[0051] A dessert connected body refers to a three-dimensional connected body in a three-dimensional NSSCI property volume that meets the lower limit of the natural dessert layer parameters, and whose internal natural permeability index NFCI, porosity Φ, and natural dessert comprehensive index NSSCI all reach the industrial lower limit value.

[0052] The following are specific examples shown in this embodiment: The research subject is the Huangliu Group of the DFX gas field.

[0053] The Huangliu Formation reservoir in the DFX gas field has extremely poor physical properties and strong heterogeneity. By using laser particle size analysis data and conventional well logging curves (GR, RT, RHOB, CNL, DT), a nonlinear fitting was performed to obtain a curve model of the median particle size and sorting coefficient.

[0054] like Figure 2 As shown, Figure 2 The curve prediction model diagram is the particle size / sorting coefficient-equivalent pore throat radius curve. Based on the Kozeny-Carman equation, the equivalent pore throat curve model is further calculated by fitting parameters and curve models from mercury intrusion porosimetry experiments on different samples.

[0055] like Figure 3 As shown, Figure 3 The image shows the Natural Permeability Index (NFCI) curve and the Natural Sweetness Composite Index (NSSCI) curve, and the Natural Permeability Index (NFCI) is calculated using an expression.

[0056] Where μ_g is the formation natural gas viscosity after PVT correction, with a value of 0.07298 mPa·s; B_g is the natural gas volume factor, with a value of 0.00627.

[0057] Using gas production capacity data from test gas analysis or core samples, determine the industrial lower limits of the natural permeability index NFCI and porosity Φ, NFCI_min and Φ_min.

[0058] Among them, NFCI_min is 3.4733; Φ_min is 12.7%. The cumulative thickness h_net of the "natural sweet spot layer" in different wells in the study area is 10.5m.

[0059] like Figure 3 As shown, the energy factor EF of different wells is calculated and combined with the natural permeability index NFCI to construct the natural sweet spot index NSSCI model.

[0060] like Figure 4 As shown, based on the NSSCI curve, a three-dimensional NSSCI attribute volume is obtained through geostatistical inversion. By setting a threshold, the attribute volume is hollowed out and sculpted to obtain a connected volume of natural desserts.

[0061] like Figure 5 As shown, the NSSCI attribute threshold is set to 44.2033. This sweet spot connected component is a high-quality reservoir 3D model that naturally possesses industrial flow capabilities.

[0062] The embodiments of the present invention have been described in detail above, but the content described is only a preferred embodiment of the present invention and should not be considered as limiting the scope of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the patent coverage of the present invention.

Claims

1. A method for predicting sweet spots of a super-pressured low-permeability gas reservoir based on a grain-size-reconstructed percolation evaluation model, characterized in that, The specific steps include: S1. Using a neural network algorithm, a multivariate nonlinear relationship is fitted based on core grain size and well logging curve data to obtain a core grain size curve prediction model. Through the core grain size curve prediction model, a grain size median curve prediction model and a sorting coefficient curve prediction model are obtained. S2. Substitute the core grain size curve prediction model, grain size median curve prediction model, and sorting coefficient curve prediction model obtained in S1 into the conversion relationship model between grain size parameters and pore throat radius. Through the conversion relationship model between grain size parameters and pore throat radius, calculate the equivalent pore throat radius curve model. S3. Define the Natural Permeability Index (NFCI) and obtain the NFCI definition formula. S4. Substitute the definition formula of the natural seepage capacity index NFCI in S3 into the equivalent pore throat radius curve model obtained in S2, calculate the natural seepage capacity index NFCI, and calculate the equivalent pore throat radius curve model as the natural seepage capacity index curve model. S5. Using gas production capacity data from gas testing or core samples, determine the industrial lower limits of the natural permeability index NFCI and porosity Φ, NFCI_min and Φ_min; on a single well profile, identify all segments that simultaneously satisfy NFCI > NFCI_min and Φ > Φ_min, define the identified segments as "natural sweet spots", and calculate the cumulative thickness h_net of the "natural sweet spots"; S6. Define the energy factor EF and obtain the formula for defining the energy factor EF; construct the NSSCI curve model of the natural dessert comprehensive index based on the formula for defining the energy factor EF. S7. Extract the NFCI values ​​from the natural permeability index curve model point by point according to depth. Based on the NFCI_min and Φ_min values ​​determined in S5, select the depth points in the "natural sweet spot layer". Substitute the selected depth points into the natural sweet spot comprehensive index NSSCI curve model to obtain the NSSCI curve model that couples permeability and formation energy properties. S8. Using the natural sweet spot index (NSSCI) curves of each single well as hard data, three-dimensional seismic attributes sensitive to porosity, permeability, and gas content are simultaneously selected as auxiliary constraint data. Geostatistical methods are used to establish a three-dimensional NSSCI attribute volume. In the three-dimensional NSSCI attribute volume, continuous spatial regions with NSSCI values ​​greater than a set threshold are extracted, and the obtained continuous spatial regions are defined as "sweet spot connected bodies".

2. The method of claim 1, wherein the method is characterized by, The conversion model between particle size parameters and pore throat radii in S2 is as follows: r_equiv = a * (D50)^b * (Sort)^c * Φ^d (1) Where a, b, c, and d are coefficients obtained by fitting mercury intrusion porosimetry data from the core, D50 is the median grain size curve, and Sort is the sorting coefficient curve.

3. The method of claim 1, wherein the method is characterized by, The specific formula for defining the Natural Permeability Index (NFCI) is as follows: NFCI = (K * r_equiv) / (μ_g * B_g) (2) Where K is the permeability calculated from the particle size or pore throat parameters, μ_g is the formation natural gas viscosity after PVT correction, and B_g is the natural gas volume factor.

4. The method for predicting sweet spots in overpressured, low-permeability gas reservoirs based on a particle size reconstruction seepage evaluation model according to claim 1, characterized in that, The energy factor EF is defined by the following formula: EF = (P_p - P_wf) / P_wf (3) Where P_p is the formation pore pressure and P_wf is the reasonable bottom flow pressure of the production well.

5. The method for predicting sweet spots in overpressured, low-permeability gas reservoirs based on a particle size reconstruction seepage evaluation model according to claim 4, characterized in that, The formula for calculating the Natural Sweetness Index (NSSCI) is as follows: NSSCI = NFCI * EF * (1 - S_wi) * h_net (4) Where S_wi is the bound water saturation and h_net is the cumulative thickness of the "natural dessert layer".

6. The method for predicting sweet spots in overpressured, low-permeability gas reservoirs based on a particle size reconstruction seepage evaluation model according to claim 1, characterized in that: S1 uses software to execute the neural network algorithm, and the execution software is PyCharm.

Images